Paper Abstract and Keywords |
Presentation |
2022-01-18 14:00
Robustness to Adversarial Examples by Mixtures of L1 Regularazation Models Hironobu Takenouchi, Junichi Takeuchi (Kyushu Univ.) IBISML2021-26 |
Abstract |
(in Japanese) |
(See Japanese page) |
(in English) |
We propose a method of adversarial training using L1 regularizationfor image classification.It is known that L1 regularization is effective to improve robustness for adversarial samples by pruning image classifier's explanatory variables.However, we showed by experiments that the regularized model is robust to the adversarial samples to the model without L1 regularization,while it has vulnerability to the adversarial samples generated using the knowledgeabout the L1 regularized model.For this problem, we developed a new method using amixture of models with various strength of L1 regularizationand trained it with adversarial samples to single L1 regularized models.By experiment, we showed that our model is more robust than the single modelsand that the successful adversarial perturbation to the mixture modelis less diverse than that to other single models. |
Keyword |
(in Japanese) |
(See Japanese page) |
(in English) |
Deep Learning / Adversarial Training / / / / / / |
Reference Info. |
IEICE Tech. Rep., vol. 121, no. 321, IBISML2021-26, pp. 61-66, Jan. 2022. |
Paper # |
IBISML2021-26 |
Date of Issue |
2022-01-10 (IBISML) |
ISSN |
Online edition: ISSN 2432-6380 |
Copyright and reproduction |
All rights are reserved and no part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopy, recording, or any information storage and retrieval system, without permission in writing from the publisher. Notwithstanding, instructors are permitted to photocopy isolated articles for noncommercial classroom use without fee. (License No.: 10GA0019/12GB0052/13GB0056/17GB0034/18GB0034) |
Download PDF |
IBISML2021-26 |
Conference Information |
Committee |
IBISML |
Conference Date |
2022-01-17 - 2022-01-18 |
Place (in Japanese) |
(See Japanese page) |
Place (in English) |
Online |
Topics (in Japanese) |
(See Japanese page) |
Topics (in English) |
Machine Learning, etc. |
Paper Information |
Registration To |
IBISML |
Conference Code |
2022-01-IBISML |
Language |
Japanese |
Title (in Japanese) |
(See Japanese page) |
Sub Title (in Japanese) |
(See Japanese page) |
Title (in English) |
Robustness to Adversarial Examples by Mixtures of L1 Regularazation Models |
Sub Title (in English) |
|
Keyword(1) |
Deep Learning |
Keyword(2) |
Adversarial Training |
Keyword(3) |
|
Keyword(4) |
|
Keyword(5) |
|
Keyword(6) |
|
Keyword(7) |
|
Keyword(8) |
|
1st Author's Name |
Hironobu Takenouchi |
1st Author's Affiliation |
Kyushu University (Kyushu Univ.) |
2nd Author's Name |
Junichi Takeuchi |
2nd Author's Affiliation |
Kyushu University (Kyushu Univ.) |
3rd Author's Name |
|
3rd Author's Affiliation |
() |
4th Author's Name |
|
4th Author's Affiliation |
() |
5th Author's Name |
|
5th Author's Affiliation |
() |
6th Author's Name |
|
6th Author's Affiliation |
() |
7th Author's Name |
|
7th Author's Affiliation |
() |
8th Author's Name |
|
8th Author's Affiliation |
() |
9th Author's Name |
|
9th Author's Affiliation |
() |
10th Author's Name |
|
10th Author's Affiliation |
() |
11th Author's Name |
|
11th Author's Affiliation |
() |
12th Author's Name |
|
12th Author's Affiliation |
() |
13th Author's Name |
|
13th Author's Affiliation |
() |
14th Author's Name |
|
14th Author's Affiliation |
() |
15th Author's Name |
|
15th Author's Affiliation |
() |
16th Author's Name |
|
16th Author's Affiliation |
() |
17th Author's Name |
|
17th Author's Affiliation |
() |
18th Author's Name |
|
18th Author's Affiliation |
() |
19th Author's Name |
|
19th Author's Affiliation |
() |
20th Author's Name |
|
20th Author's Affiliation |
() |
Speaker |
Author-1 |
Date Time |
2022-01-18 14:00:00 |
Presentation Time |
20 minutes |
Registration for |
IBISML |
Paper # |
IBISML2021-26 |
Volume (vol) |
vol.121 |
Number (no) |
no.321 |
Page |
pp.61-66 |
#Pages |
6 |
Date of Issue |
2022-01-10 (IBISML) |
|